Neural network modeling and control of proton exchange membrane fuel cell

A neural network model and fuzzy neural network controller was designed to control the inner impedance of a proton exchange membrane fuel cell (PEMFC) stack. A radial basis function (RBF) neural network model was trained by the input-output data of impedance. A fuzzy neural network controller was designed to control the impedance response. The RBF neural network model was used to test the fuzzy neural network controller. The results show that the RBF model output can imitate actual output well, the maximal error is not beyond 20 m-, the training time is about 1 s by using 20 neurons, and the mean squared errors is 141.9 m-2. The impedance of the PEMFC stack is controlled within the optimum range when the load changes, and the adjustive time is about 3 min.     

Control-oriented dynamic fuzzy model and predictive control for proton exchange membrane fuel cell stack

Proton exchange membrane fuel cell （PEMFC） stack temperature and cathode stoichiometric oxygen are very important control parameters. The performance and lifespan of PEMFC stack are greatly dependent on the parameters. So, in order to improve the performance index, tight control of two parameters within a given range and reducing their fluctuation are indispensable, However, control-oriented models and control strategies are very weak junctures in the PEMFC development. A predictive control algorithm was presented based on their model established by input-output data and operating experiences. It adjusts the operating temperature to 80℃. At the same time, the optimized region of stoichiometric oxygen is kept between 1.8-2.2. Furthermore, the control algorithm adjusts the variants quickly to the destination value and makes the fluctuation of the variants the least. According to the test results, compared with traditional fuzzy and PID controllers, the designed controller shows much better performance.     

神经网络-标称系统混合模型的离散变结构控制

提出将神经网络和标称系统混合建模方法引入到柔性结构主动控制当中，在混合模型的基础上，利用离散变结构控制（VSC）对柔性结构振动进行控制．离散变结构控制的滑模面是以标称系统为基础，由最优化二次型价值函数确定，并通过黎卡提方程求解．利用标称模型和神经网络混合建模方法来减小系统的不确定性，达到减弱变结构控制在实际控制系统中的抖动问题．神经网络采用多层前馈网络（MFNN），来对不确定部分进行建模．仿真结果表明系统振动受到了有效的控制，说明提出的神经网络变结构控制（NNVSC）方法非常有效。     

基于神经网络的建筑能耗混合预测模型

为了提升建筑能耗预测的精度、鲁棒性和泛化能力,提出树种算法（TSA）优化的径向基函数（RBF）神经网络与长短时记忆（LSTM）神经网络结合的混合预测模型. 采用基于自适应噪声的完全集成经验模态分解算法,将建筑能耗数据分解为1组本征模态函数（IMF）分量和1个残余分量,利用样本熵算法将各分量划分为高频分量和低频分量. 采用最小绝对收缩与选择算子（LASSO）方法进行特征选择. 分别利用TSA算法优化后的RBF模型与LSTM模型对低频分量和高频分量进行预测,并叠加重构得到最终预测结果. 模型评估结果表明,混合预测模型的精度为98. 72%. 相比于RBF、TSA-RBF、LSTM模型,所提模型的预测效果更好,且具有较强的鲁棒性和泛化能力,能够更为有效地用于建筑逐时电力能耗预测.     

基于混合神经网络的风机性能监测模型

针对传统的RBF神经网络泛化能力差的缺点,利用RBF神经网络强大的非线性逼近能力和数学模型良好的外推能力,提出了一种将传统的RBF神经网络和用偏最小二乘法建立的通风机性能数学模型相结合的混合神经网络模型,并将该模型用于通风机的重要性能参数——流量的监测上。以实验室4-73No.8D离心风机为研究对象,用不同导流器开度下的实验数据进行拟合,研究结果表明,混合神经网络模型的泛化能力强,精度高,各项模型评价参数均优于传统的RBF神经网络模型。     

Estimation of equivalent internal-resistance of PEM fuel cell using artificial neural networks

A practical method of estimation for the internal-resistance of polymer electrolyte membrane fuel cell (PEMFC) stack was adopted based on radial basis function (RBF) neural networks. In the training process, k-means clustering algorithm was applied to select the network centers of the input training data. Furthermore, an equivalent electrical-circuit model with this internal-resistance was developed for investigation on the stack. Finally using the neural networks model of the equivalent resistance in the PEMFC stack, the simulation results of the estimation of equivalent internal-resistance of PEMFC were presented. The results show that this electrical PEMFC model is effective and is suitable for the study of control scheme, fault detection and the engineering analysis of electrical circuits.     

基于六维前馈神经网络模型的图像增强算法

针对滤波去噪对边缘造成的弱化、部分采集图像不清晰以及对比度低的问题,在充分分析模型的动力学性质的基础上,提出一种基于六维前馈神经网络模型的图像增强算法。试验表明:基于六维前馈神经网络模型的图像增强算法可以更好地达到图像增强效果。与其它几种增强算法相比,增强效果清晰,且算法更优。     

动态递归模糊神经网络及其BP学习算法

提出了一种新型的动态递归模糊神经网络，并根据动态递归神经网络的数学模型推导出其动态反向传播学习算法，仿真结果表明对于动态系统的辨识，动态递归模糊神经网络较传统模糊神经网络在辨识精度和稳定性方面具有更好的效果。     

不同流道结构质子交换膜燃料电池内传递现象的三维模拟

应用计算流体力学方法，建立了用于模拟质子交换膜燃料电池（PEMFC）传递特性和电化学性能的稳态、等温的三维数学模型。计算了传统流道和交叉梳状流道燃料电池的流场、电流密度和组分浓度等的多维分布。与传统流道的燃料电池相比，交叉梳状流道所产生的电极内强烈的强制对流机理提高了反应物和产物的传输速率，从而改善了电池的极限电流和极化性能等。利用模型估算的极化特性和文献实验结果吻合较好。     

Variable cluster analysis method for building neural network model

To address the problems that input variables should be reduced as much as possible and explain output variables fully in building neural network model of complicated system, a variable selection method based on cluster analysis was investigated. Similarity coefficient which describes the mutual relation of variables was defined. The methods of the highest contribution rate, part replacing whole and variable replacement are put forwarded and de-duced by information theory. The software of tbe neural network based on cluster analysis, whicb can provide many kinds of metbods for defining variable similarity coefficient, clustering system variable and evaluating variable clus-ter, was developed and applied to build neural network forecast model of cement clinker quality. Tile results show that all the network scale, training time and prediction accuracy are perfect. The practical application demonstrates that the method of selecting variables for neural network is feasible and effective.     

基于BP神经网络和多元Taylor级数的混合定位算法

针对多元Taylor级数算法定位精度严重依赖初始值的问题,提出一种新的混合定位算法。通过BP神经网络定位算法提供初始值,提高多元Taylor级数展开法的收敛速度;通过多元Taylor级数展开法,充分利用未知节点之间的距离信息,减小测距误差造成的定位误差。仿真结果表明:混合定位算法的精度更高,并且减少了网格间距对定位精度的影响。     

时延动态神经网络的稳定性分析

基于时延动态网络模型的优化计算易于收敛到非法解或局部极小解,以及其算法对模型参数和初始条件具有很强的依赖性等缺点,对时延动态神经网络的稳定性进行了深入地研究讨论,通过Razumikhin-type定理对时延动态神经网络的稳定性进行了分析研究.     

基于动态模糊神经网络的产品成本估算

针对在产品方案设计阶段成本估算信息少且颗粒度大的问题，结合神经网络和模糊工程技术提出了动态模糊神经网络（DFNN），采用模糊推理的信息处理方法，在学习过程中隐层层数及维数根据规则不断变化，神经网络结构呈现动态.研究了动态模糊神经网络的学习过程、网络动态算法及模糊知识处理方法，建立了成本估算模型，并开发了基于动态模糊神经网络的成本估算软件，实现了利用产品方案设计信息自动进行成本估算.以挖掘机和液压油缸为例进行验证，结果表明该方法具有较强的信息处理能力，并提高了成本估算模型的柔性.     

Identification and analysis based on genetic algorithm for proton exchange membrane fuel cell stack

The temperature of proton exchange membrane fuel cell stack and the stoichiometric oxygen in cathode have relationship with the performance and life span of fuel cells closely. The thermal coefficients were taken as important factors affecting the temperature distribution of fuel cells and components. According to the experimental analysis, when the stoichiometric oxygen in cathode is greater than or equal to 1.8, the stack voltage loss is the least. A novel genetic algorithm was developed to identify and optimize the variables in dynamic thermal model of proton exchange membrane fuel cell stack, making the outputs of temperature model approximate to the actual temperature, and ensuring that the maximal error is less than 1℃. At the same time, the optimum region of stoichiometric oxygen is obtained, which is in the range of 1.8 -2.2 and accords with the experimental analysis results. The simulation and experimental results show the effectiveness of the proposed algorithm.     

基于现场数据的电站过热器神经网络建模

为克服机理建模方法的不足,对利用现场数据建立过热汽温模型进行了研究。通过对过热器运行机理的分析,确定了影响过热汽温变化的主要因素,及其对过热汽温影响的延迟时间。在论述了神经网络建模原理的基础上,建立神经网络模型。通过大量现场数据的训练,使所建模的输出与实际系统的输出基本吻合。最后,用部分现场数据对所建模型进行了仿真试验,证明了该建模方法的可行性。基于神经网络建模计算速度快及模型精度高,模型输出基本上反映了过热汽温的实际运行情况。     

大柔性变结构系统传输特性的反向传播网络辨识

对大射电望远镜光机电一体化设计中悬索和馈源舱这一大柔性变结构系统的传输特性进行了研究 .由于该系统的结构是动态变化的 ,采用常规方法无法确定其传输特性 ,因此用改进的反向传播算法对其进行了辨识 ,得到了系统的反向传播网络模型 .从测试数据的网络输出结果与理论计算结果的对比可以看出 ,该网络模型能够较好地反映这一系统的输入 /输出特性 .这一结果对大射电望远镜系统的整体建模和馈源舱位置的间接测量都具有意义 ,也为柔性结构系统传输特性的研究提供了一种可资借鉴的方法 .     

神经网络在水质模型中的应用

将神经网络理论和算法应用于水环境评价中，基于RBF神经网络进行水质建模，应用国内河流的实测样本对模型进行训练和检验．该方法较之原有的水质数学模型以及其他全局逼近神经网络方法的结果更加精确，适用范围更广．     

基于PID神经网络的拉深件变压边力控制方法研究

将传统的PID控制器与神经元网络融合,建立PID神经网络。运用VB语言编译了拉深过程中变压边力控制系统的PID神经网络仿真程序,并与传统的PID控制仿真进行了对比。建立了基于PIDNN的变压边力控制系统,并通过锥形件拉深实验,证实了PIDNN控制系统具有精度高,抗干扰能力强,能较准确达到变压边力控制要求等优越性。为后续的变压边力控制系统实验研究与工厂实际应用提供了理论与实践基础。